@openclaw/voice-call 2026.9.5 → 2026.9.7

package/dist/.setup/{realtime-handler-CaoetLrM.mjs → realtime-handler-BgDPcPZt.mjs}

@@ -1,7 +1,7 @@
-import { d as REALTIME_VOICE_END_CALL_TOOL_NAME, s as resolveVoiceCallPublicPathPrefix } from "./config-72FnMWHb.mjs";
-import { c as canonicalizeVoiceCallMediaBase64, o as WebSocket, s as WebSocketServer } from "./runtime-entry-YWSSnAsH.mjs";
+import { d as REALTIME_VOICE_END_CALL_TOOL_NAME, s as resolveVoiceCallPublicPathPrefix } from "./config-Ds7xoZbI.mjs";
+import { c as canonicalizeVoiceCallMediaBase64, o as WebSocket, s as WebSocketServer } from "./runtime-entry-BNOQRXii.mjs";
 import { formatErrorMessage } from "openclaw/plugin-sdk/error-runtime";
-import { asNullableRecord, asOptionalObjectRecord } from "openclaw/plugin-sdk/string-coerce-runtime";
+import { asNullableRecord, asOptionalObjectRecord, asOptionalRecord, normalizeOptionalString } from "openclaw/plugin-sdk/string-coerce-runtime";
 import { createSubsystemLogger } from "openclaw/plugin-sdk/runtime-env";
 import { REALTIME_VOICE_AGENT_CONSULT_TOOL_NAME, REALTIME_VOICE_AUDIO_FORMAT_G711_ULAW_8KHZ, buildRealtimeVoiceAgentConsultWorkingResponse, buildRealtimeVoiceAgentErrorProviderResult, calculateMulawRms, createRealtimeVoiceSessionHarness, createSpeechThresholdGate, readRealtimeVoiceConsultQuestion, readSpeakableRealtimeVoiceToolResult, resolveRealtimeVoiceBargeIn, resolveRealtimeVoiceSessionPolicy } from "openclaw/plugin-sdk/realtime-voice";
 import { normalizeWebhookPath } from "openclaw/plugin-sdk/webhook-ingress";
@@ -266,13 +266,9 @@ function tryParseJson(rawMessage) {
 	} catch {}
 	return null;
 }
-/** Read an object-valued field from a parsed frame. */
-function readRecordField(record, field) {
-	return asOptionalObjectRecord(record[field]);
-}
 /** Parse a common provider media frame. */
 function parseMediaFrame(msg) {
-	const mediaData = readRecordField(msg, "media");
+	const mediaData = asOptionalObjectRecord(msg.media);
 	const payload = typeof mediaData?.payload === "string" ? mediaData.payload : void 0;
 	const canonicalPayload = payload ? canonicalizeVoiceCallMediaBase64(payload) : void 0;
 	if (!canonicalPayload) return { kind: "ignored" };
@@ -285,35 +281,30 @@ function parseMediaFrame(msg) {
 }
 /** Parse a common provider mark frame. */
 function parseMarkFrame(msg) {
-	const markData = readRecordField(msg, "mark");
+	const markData = asOptionalObjectRecord(msg.mark);
 	return {
 		kind: "mark",
 		name: typeof markData?.name === "string" ? markData.name : void 0
 	};
 }
-/** Parse common media, mark, and stop frames shared by supported providers. */
-function parseCommonInboundFrame(event, msg) {
-	if (event === "media") return parseMediaFrame(msg);
-	if (event === "mark") return parseMarkFrame(msg);
-	if (event === "stop") return { kind: "stop" };
-}
 /** Parse one provider frame with provider-specific start/error hooks. */
 function parseProviderInboundFrame(rawMessage, parseStartFrame, parseExtraFrame) {
 	const msg = tryParseJson(rawMessage);
 	if (!msg) return { kind: "ignored" };
 	const event = msg.event;
-	if (event === "start") return parseStartFrame(msg) ?? { kind: "ignored" };
-	return parseCommonInboundFrame(event, msg) ?? parseExtraFrame?.(event, msg) ?? { kind: "ignored" };
-}
-/** Include streamSid only when Twilio has already supplied one. */
-function withOptionalStreamSid(streamSid) {
-	return streamSid === void 0 ? {} : { streamSid };
+	switch (event) {
+		case "start": return parseStartFrame(msg) ?? { kind: "ignored" };
+		case "media": return parseMediaFrame(msg);
+		case "mark": return parseMarkFrame(msg);
+		case "stop": return { kind: "stop" };
+		default: return parseExtraFrame?.(event, msg) ?? { kind: "ignored" };
+	}
 }
 /** Serialize a provider media frame. */
 function serializeMediaFrame(payloadBase64, streamSid) {
 	return JSON.stringify({
 		event: "media",
-		...withOptionalStreamSid(streamSid),
+		streamSid,
 		media: { payload: payloadBase64 }
 	});
 }
@@ -321,14 +312,14 @@ function serializeMediaFrame(payloadBase64, streamSid) {
 function serializeClearFrame(streamSid) {
 	return JSON.stringify({
 		event: "clear",
-		...withOptionalStreamSid(streamSid)
+		streamSid
 	});
 }
 /** Serialize a provider mark frame. */
 function serializeMarkFrame(name, streamSid) {
 	return JSON.stringify({
 		event: "mark",
-		...withOptionalStreamSid(streamSid),
+		streamSid,
 		mark: { name }
 	});
 }
@@ -341,7 +332,7 @@ var TwilioStreamFrameAdapter = class {
 	/** Parse one Twilio websocket message into a normalized frame. */
 	parseInbound(rawMessage) {
 		return parseProviderInboundFrame(rawMessage, (msg) => {
-			const startData = readRecordField(msg, "start");
+			const startData = asOptionalObjectRecord(msg.start);
 			const streamSid = typeof startData?.streamSid === "string" ? startData.streamSid : "";
 			const callSid = typeof startData?.callSid === "string" ? startData.callSid : "";
 			if (!streamSid || !callSid) return;
@@ -375,7 +366,7 @@ var TelnyxStreamFrameAdapter = class {
 	parseInbound(rawMessage) {
 		return parseProviderInboundFrame(rawMessage, (msg) => {
 			const topLevelStreamId = typeof msg.stream_id === "string" && msg.stream_id ? msg.stream_id : void 0;
-			const startData = readRecordField(msg, "start");
+			const startData = asOptionalObjectRecord(msg.start);
 			const providerCallId = typeof startData?.call_control_id === "string" && startData.call_control_id ? startData.call_control_id : void 0;
 			if (!topLevelStreamId || !providerCallId) return;
 			return {
@@ -385,7 +376,7 @@ var TelnyxStreamFrameAdapter = class {
 			};
 		}, (event, msg) => {
 			if (event !== "error") return;
-			const errorData = readRecordField(msg, "payload");
+			const errorData = asOptionalObjectRecord(msg.payload);
 			return {
 				kind: "error",
 				code: typeof errorData?.code === "string" || typeof errorData?.code === "number" ? String(errorData.code) : void 0,
@@ -431,14 +422,6 @@ function buildGreetingInstructions(baseInstructions, greeting) {
 	const intro = "Start the call by greeting the caller naturally. Include this greeting in your first spoken reply:";
 	return baseInstructions ? `${baseInstructions}\n\n${intro} "${trimmedGreeting}"` : `${intro} "${trimmedGreeting}"`;
 }
-function readConsultArgText(args, key) {
-	if (!args || typeof args !== "object" || Array.isArray(args)) return;
-	const value = args[key];
-	return typeof value === "string" && value.trim() ? value.trim() : void 0;
-}
-function readConsultQuestionText(args) {
-	return readRealtimeVoiceConsultQuestion(args);
-}
 function normalizeTranscriptText(text) {
 	return text.replace(/\s+/g, " ").trim();
 }
@@ -487,11 +470,11 @@ function limitPartialUserTranscript(text) {
 	return tail.replace(/^\S+\s+/, "").trimStart() || tail.trimStart();
 }
 function withFallbackConsultQuestion(args, fallback) {
-	const providerQuestion = readConsultQuestionText(args);
+	const providerQuestion = readRealtimeVoiceConsultQuestion(args);
 	const question = fallback?.trim();
 	if (providerQuestion) {
 		if (question && providerQuestion.length <= 40 && question.length >= providerQuestion.length + 8) {
-			const context = readConsultArgText(args, "context");
+			const context = normalizeOptionalString(asOptionalRecord(args)?.context);
 			const fallbackContext = `Realtime provider supplied a shorter consult question: ${providerQuestion}`;
 			return args && typeof args === "object" && !Array.isArray(args) ? {
 				...args,
@@ -1702,25 +1685,23 @@ var RealtimeCallHandler = class {
 			...callerMeta.from ? { from: callerMeta.from } : {},
 			...callerMeta.to ? { to: callerMeta.to } : {}
 		};
-		const callRecord = await this.resolveRealtimeCall(callSid, callerMeta, baseFields);
-		if (!callRecord) return null;
-		return {
-			callRecord,
-			baseFields
-		};
-	}
-	async resolveRealtimeCall(callSid, callerMeta, baseFields) {
+		let callRecord;
 		if (callerMeta.callId) {
 			const call = await this.manager.getCallForStream(callerMeta.callId);
-			return call?.providerCallId === callSid ? call : null;
+			callRecord = call?.providerCallId === callSid ? call : void 0;
+		} else {
+			await this.manager.processEvent({
+				id: `realtime-initiated-${callSid}`,
+				callId: callSid,
+				type: "call.initiated",
+				...baseFields
+			});
+			callRecord = this.manager.getCallByProviderCallId(callSid);
 		}
-		await this.manager.processEvent({
-			id: `realtime-initiated-${callSid}`,
-			callId: callSid,
-			type: "call.initiated",
-			...baseFields
-		});
-		return this.manager.getCallByProviderCallId(callSid) ?? null;
+		return callRecord ? {
+			callRecord,
+			baseFields
+		} : null;
 	}
 	async executeEndCallTool(params) {
 		const binding = this.activeTelephonyBindingsByCallId.get(params.callId);

package/dist/.setup/{response-generator-DB1nElLO.mjs → response-generator-BzKrJMEb.mjs}

@@ -1,5 +1,5 @@
-import { c as resolveVoiceCallSessionKey } from "./config-72FnMWHb.mjs";
-import { _ as resolveCallAgentId, g as resolveVoiceResponseModel } from "./runtime-entry-YWSSnAsH.mjs";
+import { c as resolveVoiceCallSessionKey } from "./config-Ds7xoZbI.mjs";
+import { _ as resolveCallAgentId, g as resolveVoiceResponseModel } from "./runtime-entry-BNOQRXii.mjs";
 import { filterStringEntries, isRecord, normalizeLowercaseStringOrEmpty, normalizeStringEntries } from "openclaw/plugin-sdk/string-coerce-runtime";
 import { resolveAgentConfig } from "openclaw/plugin-sdk/agent-scope-runtime";
 import { isValidAgentHarnessSessionStoreEntry } from "openclaw/plugin-sdk/session-store-runtime";
@@ -282,6 +282,7 @@ async function generateVoiceResponse(params) {
 				toolsAllow,
 				abortSignal,
 				blockReplyBreak: "text_end",
+				resolveReplyDelivery: async (minimumAssistantMessageIndex = 0) => deliveredEarly && minimumAssistantMessageIndex === 0 ? "pending" : "missing",
 				onBlockReply: (payload, context) => {
 					if (latestToolBoundaryMessageIndex !== void 0) {
 						const messageIndex = context?.assistantMessageIndex;